Call Centers: A New Paradigm

Last year I wrote a blog post summarizing studies made on call centers. It covered many aspects of contact centers from how good customer service improves customer loyalty and company growth, to the anxiety and stress factors experienced by call center workers.

Today we'll explore the opportunities call centers have to change the old "people taking phone calls" paradigm for a new one, without leaving people on the sidelines.

Towards Enhancement not Automation

The current trend in cutting-edge call center automation is that AI agents are pretty good but still require supervision. What today's Virtual Call Center Platforms do is to enable companies to assist live agents and magnify their productivity.

Agent Assistance

Agent assistance is a way to empower Human agents with continuous support during their calls by listening to the customer's audio stream, identifying intent and displaying step-by-step suggestions for live agents. Agent Assistance is a very useful tool for call centers as it allows companies to standardize procedures for a more uniform customer experience as well as keep track of customer preferences and behavior by listening for key-words or phrases.

Job Description: AI Supervisor

The basic idea is 1 Human being supervising several AI Systems. If we assume this theoretical speaking AI System has a success rate of 75% (i.e. it's capable of flawlessly completing ~7 calls out of 10), there are still times when the Human supervisor will need to get involved.

To get an idea of what this new job might entail, we can view the performance of the cluster of AI Systems as a binomial distribution in which at any point in time an AI System will either successfully complete a call or fail to do so and force the AI Supervisor to take over :

We can have a look at what this would mean for different configurations:

TL;DR Supervising 4 bots maximizes the Human agent's productivity and minimizes the chance that they need to intervene on 2 calls "at once". Supervising more than 4 bots will lead to Human supervisors being overwhelmed and longer call handling times for clients.

To be fair, a 75% success rate is fairly low and depending on the level of sophistication of the A.I. and the complexity of the conversations, supervising 6 or more bots is probably feasible in the near-term.

--- Please log in to read more ---